Currently exclusive to NVIDIA Turing GPUs, Variable Rate Shading allows shader resources to be dynamically allocated to sections of a scene depending on complexity and other context clues.
Currently exclusive to NVIDIA Turing GPUs, Variable Rate Shading allows shader resources to be dynamically allocated to sections of a scene depending on complexity and other context clues.
NVIDIA’s Turing architecture is currently extremely new and broadly unexplored in terms of its capabilities. RTX-specific features have been promoted extensively but the broader set available to both GeForce RTX and GTX cards remain only gently touched upon. One class of features in particular – Advanced Shading – has had its potential only hinted at with a retrofit into Wolfenstein II: The New Colossus in the form of Variable Rate Shading, and now it appears that AMD have their own form in the works.
As reported today by PCGamesN an intrepid twitter user going by the handle @0x22h discovered a patent for ‘Variable Rate Shading’ filed by AMD in 2017. Patent 15/687421 (Application Number) describes Variable Rate Shading methodology outlined as:
“A technique for performing rasterization and pixel shading with decoupled resolution is provided herein. The technique involves performing rasterization as normal to generate fine rasterization data and a set of (fine) quads. The quads are accumulated into a tile buffer and coarse quads are generated from the quads in the tile buffer based on a shading rate. The shading rate determines how many pixels of the fine quads are combined to generate coarse pixels of the coarse quads. Combination of fine pixels involves generating a single coarse pixel for each such fine pixel to be combined. The positions of the coarse pixels of the coarse quads are set based on the positions of the corresponding fine pixels. The coarse quads are shaded normally and the resulting shaded coarse quads are modified based on the fine rasterization data to generate shaded fine quads. “
The exact process is likely to differ from NVIDIA’s implementation but the intent is very similar: by analysing context clues and frame-by-frame differences, adjust the shading rate of sections of the scene and through more optimal allocation of shader resources improve overall GPU performance. In briefings ahead of the GeForce GTX 1660 Ti’s launch NVIDIA stated that Variable Rate Shading contributed to an estimated 50% performance improvement for the card over the GTX 1060 6GB, and that’s a game which may not be particularly well suited to the task.
PCGamesN rightly state that this has significant implications for both AMD-powered PC gaming and consoles. Keeping up with NVIDIA technologically is always going to be a factor of high importance to AMD, but the Red Team also have a stranglehold on a console space that can’t simple have a larger GPU thrown at it when more demanding games are released. Niche rendering techniques of today can be exploited heavily when the hardware involved is fixed, leading to meaningful performance gains in the future.
It remains to be seen whether the Patent will be upheld, but this is just the sort of technology one would expect to see supported in AMD’s next-generation GPU architecture, beginning with NAVI later this year.
SOURCE: PCGamesN, FreePatentsOnline.com
